2.6.0-test1-ac2 issues / Toshiba Laptop keyboard

Finally, I was able to get 2.6.0-test1-ac2 working.

Some issues I found:

* eepro100 is b0rked:

eepro100: Unknown symbol mii_ethtool_sset
eepro100: Unknown symbol mii_link_ok
eepro100: Unknown symbol mii_check_link
eepro100: Unknown symbol mii_nway_restart
eepro100: Unknown symbol mii_ethtool_gset

but by switchin to e100 instead, I got my NIC working

* The kernel reports itself as "Linux version 2.6.0-test1-ac1" but IS ac2!

* The IDE ATA disk works, but upon reboot, the machine does NOT find the IDE harddisk anymore! Tis means I have to turn the machine off and on again (since it has no reset button)

Details:

Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
ICH3M: IDE controller at PCI slot 0000:00:1f.1
ICH3M: chipset revision 2
ICH3M: not 100% native mode: will probe irqs later

    ide0: BM-DMA at 0xcfa0-0xcfa7, BIOS settings: hda:DMA, hdb:pio
    ide1: BM-DMA at 0xcfa8-0xcfaf, BIOS settings: hdc:DMA, hdd:pio

hda: TOSHIBA MK4019GAX, ATA DISK drive anticipatory scheduling elevator
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: TOSHIBA DVD-ROM SD-R2102, ATAPI CD/DVD-ROM drive ide1 at 0x170-0x177,0x376 on irq 15
hda: max request size: 128KiB
hda: host protected area => 1
hda: 78140160 sectors (40008 MB), CHS=77520/16/63, UDMA(100)
/dev/ide/host0/bus0/target0/lun0: p1 p2 p3
hdc: ATAPI 24X DVD-ROM CD-R/RW drive, 2048kB Cache, UDMA(33)

* The module "sg" doesn't work:

sg: Unknown symbol scsi_do_req
sg: Unknown symbol scsi_device_get
sg: Unknown symbol scsi_block_when_processing_errors
sg: Unknown symbol scsi_ioctl
sg: Unknown symbol scsi_device_put
sg: Unknown symbol scsi_sleep
sg: Unknown symbol scsi_reset_provider
sg: Unknown symbol scsi_register_interface
sg: Unknown symbol scsi_ioctl_send_command
sg: Unknown symbol scsi_release_request
sg: Unknown symbol scsi_allocate_request
sg: Unknown symbol proc_scsi
sg: Unknown symbol print_req_sense

* ide_scsi doesn't work:

ide_scsi: Unknown symbol scsi_remove_host
ide_scsi: Unknown symbol scsi_host_put
ide_scsi: Unknown symbol scsi_sleep
ide_scsi: Unknown symbol scsi_add_host
ide_scsi: Unknown symbol scsi_adjust_queue_depth
ide_scsi: Unknown symbol scsi_host_alloc

* usb_storage barfs:

usb_storage: Unknown symbol scsi_remove_host
usb_storage: Unknown symbol scsi_host_put
usb_storage: Unknown symbol scsi_add_host
usb_storage: Unknown symbol scsi_host_alloc

* and the most interesting issue is related to the keyboard of this Toshiba laptop (Satellite Pro 6100):

The Kernel reports:

atkbd.c: Unknown key (set 2, scancode 0xb2, on isa0060/serio0) pressed. atkbd.c: Unknown key (set 2, scancode 0xae, on isa0060/serio0) pressed. atkbd.c: Unknown key (set 2, scancode 0xb1, on isa0060/serio0) pressed. atkbd.c: Unknown key (set 2, scancode 0x97, on isa0060/serio0) pressed. atkbd.c: Unknown key (set 2, scancode 0xa2, on isa0060/serio0) pressed. atkbd.c: Unknown key (set 2, scancode 0x92, on isa0060/serio0) pressed.

when I type. Back in 2.4.x, I submitted a patch to AC that fixed an issue with the VERY SAME Laptop and bouncing keys. This patch was included into 2.4.x sometime (keyboard.c). Maybe it's realted to the strange kernel message I'm seeing now?

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

This is due to Alan Cox forgetting to increment the number; He is already aware of it.

Med Venlig Hilsen/Best Regards/Mit freundlichen Grüßen
Michael Kristensen <[email protected]>

* Michael Kristensen <[email protected]>:

I thought so. This happens all the time, but I thought I'd mention it.

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

You must load mii as well. The module tools should be doing that if you are using modprobe

Yep

Curious. Could be the BIOS doesn't know how to do hard disk power management especially if its quite an old PC

* Alan Cox <[email protected]>:

I see.

It's last year's model. Not quite old I'd say.

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

I have the same problem here with a toshiba satellite (with 2.6.0-test1). It boots ok, then when I reboot it stops before loading lilo (pure blank screen with only cursor on it). If I switch off/on it goes OK.

This wasn't happening in 2.5.66. ACPI S3 suspend/resume is working fine though. Could this be related to ACPI in any way?

Álvaro Lopes

* Alvaro Lopes <[email protected]>:

Exactly the same here. Note: Alan, you already had to work around one issue with that controller in the Toshiba notebooks; they wouldn't find the disks after you added the new IDE driveres back in 2.4.20-acX.

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

Ralf> * and the most interesting issue is related to the keyboard of this
Ralf> Toshiba laptop (Satellite Pro 6100):

Ralf> The Kernel reports:

Ralf> atkbd.c: Unknown key (set 2, scancode 0xb2, on isa0060/serio0) pressed. Ralf> atkbd.c: Unknown key (set 2, scancode 0xae, on isa0060/serio0) pressed. Ralf> atkbd.c: Unknown key (set 2, scancode 0xb1, on isa0060/serio0) pressed. Ralf> atkbd.c: Unknown key (set 2, scancode 0x97, on isa0060/serio0) pressed. Ralf> atkbd.c: Unknown key (set 2, scancode 0xa2, on isa0060/serio0) pressed. Ralf> atkbd.c: Unknown key (set 2, scancode 0x92, on isa0060/serio0) pressed.

I've been hacking through a similar issue on some Dell laptops.

You need to add entries to the atkbd_set2_keycode[] array in drivers/input/keyboard/atkbd.h. Look at the #defines in inlucde/linux/input.h for stuff that matches the keys that five the unknown key printk()s, and put those values in the Nth entry of the array, where N is the scancode reported in the printk().

Eg, if the first key mentioned above were a VolumeUp key, you would want to add the value 115 to the 0xb2'th entry in the array.

Once you do that, you can run xev(1x) to see what X keycode they get, look in xkb/keycodes/xfree86 for the symbol associated with each keycode, and in xkb/symbols/* for something that associates that xkb symbol with an X11 keysym.

-JimC

* James H. Cloos Jr. <[email protected]>:

But this happened while typing NORMALLY, with no frills :) I mean, I was just typing in some unix commands - so I never even came close to the keys I never use anyway...

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

I noticed the same oddity on my Toshiba Sattelite Pro 6100 and choosed to silece it with the following (trivial) patch. No side effects so far..

Cioby

@@ -191,8 +191,8 @@
 		case ATKBD_KEY_NULL:
 			break;
 		case ATKBD_KEY_UNKNOWN:
-			printk(KERN_WARNING "atkbd.c: Unknown key (set %d, scancode %#x, on %s) %s.\n",
-				atkbd->set, code, serio->phys, atkbd->release ? "released" : "pressed");
+/*			printk(KERN_WARNING "atkbd.c: Unknown key (set %d, scancode %#x, on %s) %s.\n", 
+				atkbd->set, code, serio->phys, atkbd->release ? "released" : "pressed"); */
 			break;
 		default:
 			input_regs(&atkbd->dev, regs);

* Dumitru Ciobarcianu <[email protected]>:

That removes the symptom, but not the cause.

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

I believe this is caused by hardware itself. The key bounces and the keyboard hw generates spurious keypressings.

Álvaro

* Alvaro Lopes <[email protected]>:

Nope. This happens with all laptops of that series. We have 4 of them here.

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

Ralf> atkbd.c: Unknown key (set 2, scancode 0xb2, on isa0060/serio0) pressed.

Ralf> But this happened while typing NORMALLY, with no frills :) I Ralf> mean, I was just typing in some unix commands - so I never even Ralf> came close to the keys I never use anyway...

Interesting.

The one example I quoted above, atkbd_set2_keycode[0xb2] is in fact 0.

Perhaps your kb is doing something unusual w/ the key release events.

If the kb is, eg, |=ing w/ 0x80 rather than preceding w/ 0xf0 then 0xb2 would be the release event for 0x32. atkbd_set2_keycode[0x32] is 48 (KEY_B in input.h).

So if you get unknown scancode 0xb2 from hitting the B key, that is the cause.

Otherwise, perhaps a list of some keys and what they printk() might help debug it.

-JimC

* James H. Cloos Jr. <[email protected]>:

I would think so!

The patch that fixed the issue with the key release events was this - I submitted it to Alan who then put it into ac and then it went into 2.4.x...

Details: http://www.informatik.uni-freiburg.de/~leibl/lol/

Patch excerpt:

+ /* The following 'if' is a workaround for hardware
+ * which sometimes send the key release event twice */
+ unsigned char next_scancode = scancode|up_flag;
+ if (up_flag && next_scancode==prev_scancode) {
+ /* unexpected 2nd release event */
+ } else {
+ prev_scancode=next_scancode;
+ put_queue(next_scancode);
+ }

I haven't checked if 2.6.0 already has this!

Ralf Hildebrandt (Im Auftrag des Referat V a)   [email protected]
Charite Campus Mitte                            Tel.  +49 (0)30-450 570-155
Referat V a - Kommunikationsnetze -             Fax.  +49 (0)30-450 570-916
AIM: ralfpostfix

This patch went into 2.4.21, and we immediately got a bug report! See http://www.ussg.iu.edu/hypermail/linux/kernel/0306.1/1871.html

I wrote another patch (better, I hope), which is now in 2.4.22-pre6-ac1. It is included below. I would appreciate it if it got some testing on both Toshiba laptops and other machines.

I am also interested in helping fix it in 2.6, but let's see if the fix for 2.4 works first.

Chris

@@ -198,6 +198,7 @@
 	unsigned char keycode;
 	char up_flag = down ? 0 : 0200;
 	char raw_mode;
+	char have_keycode;
 
 	pm_access(pm_kbd);
 	add_keyboard_randomness(scancode | up_flag);

@@ -214,16 +215,30 @@
 		tty = NULL;
 	}
 	kbd = kbd_table + fg_console;
-	if ((raw_mode = (kbd->kbdmode == VC_RAW))) {
+	/*
+	 *  Convert scancode to keycode
+	 */
+	raw_mode = (kbd->kbdmode == VC_RAW);
+	have_keycode = kbd_translate(scancode, &keycode, raw_mode);
+	if (raw_mode) {
 		/*
 		 *	The following is a workaround for hardware
 		 *	which sometimes send the key release event twice 
 		 */
 		unsigned char next_scancode = scancode|up_flag;
-		if (up_flag && next_scancode==prev_scancode) {
+		if (have_keycode && up_flag && next_scancode==prev_scancode) {
 			/* unexpected 2nd release event */
 		} else {
-			prev_scancode=next_scancode;
+			/* 
+			 * Only save previous scancode if it was a key-up
+			 * and had a single-byte scancode.  
+			 */
+			if (!have_keycode)
+				prev_scancode = 1;
+			else if (!up_flag || prev_scancode == 1)
+				prev_scancode = 0;
+			else
+				prev_scancode = next_scancode;
 			put_queue(next_scancode);
 		}
 		/* we do not return yet, because we want to maintain

@@ -231,10 +246,7 @@
 		   values when finishing RAW mode or when changing VT's */
 	}
 
-	/*
-	 *  Convert scancode to keycode
-	 */
-	if (!kbd_translate(scancode, &keycode, raw_mode))
+	if (!have_keycode)
 		goto out;
 
 	/*